Anthropic embeds invisible watermarks in AI-generated text to meet EU rules
Anthropic is introducing invisible watermarks in text generated by its AI models to comply with the EU AI Act’s transparency requirements. The move aims to make AI-generated content detectable by automated systems, even if it remains unnoticeable to readers. This follows a broader industry push toward clearer provenance for synthetic media.
Editor, Lazyfounder

TABLE OF CONTENTS
Why Anthropic is adding watermarks
Anthropic, the company behind the Claude AI models, has announced plans to embed invisible watermarks in text generated by its systems. The move is a response to the European Union’s AI Act Transparency Code, which took effect on 2 August. The code requires AI-generated or AI-edited content to be marked in a way that automated systems can detect.
According to YourStory, the watermarking requirement applies to models released after the code’s effective date. Anthropic also plans to extend the technology to older models.
How the watermark works
The watermark is not visible to readers and does not appear as a label or disclaimer within the text. Instead, it is embedded directly into the generated output, according to Anthropic. The company states that the watermark can persist even when text is copied, pasted, or lightly edited. It is applied at the model level, meaning it will cover multiple Claude products, including the Claude platform API, Claude Code, Claude Cowork, and Claude Tag.
For digital media files, Anthropic is using the C2PA open standard, which attaches credentials to files to indicate their origin and whether they have been created or modified using AI.
Implications for users and publishers
For most users, the change will not alter the visible experience of interacting with Claude’s outputs. The watermark becomes detectable only when the content is scanned by a compatible detection system. This could help publishers, businesses, schools, and online platforms verify whether text originated from an AI model.
However, the effectiveness of the watermark depends on how well it survives editing. Heavy rewriting or paraphrasing could weaken or remove the signal, making detection unreliable in some cases.
Broader industry shift
Anthropic is not alone in adopting transparency measures. YourStory reports that companies like Google, Meta, Microsoft, OpenAI, Synthesia, and Black Forest Labs have also agreed to comply with the EU’s transparency framework. The industry-wide push reflects concerns about misinformation, automated publishing, and the challenges of identifying synthetic media.
Watermarking is one tool among many, including content credentials and disclosure requirements, aimed at addressing these challenges.
What this means
Lazyfounder analysis — our interpretation, not reported fact.
Anthropic’s decision to embed invisible watermarks in AI-generated text signals a shift toward built-in compliance with regulatory frameworks like the EU AI Act. For founders and operators, this move sets a precedent: transparency requirements are no longer just about disclosure but about technical integration into AI outputs.
This could accelerate the adoption of similar measures globally, especially in markets with strict content authenticity rules. Startups building on or competing with Claude may need to account for watermarking in their own models or risk falling behind on compliance. Meanwhile, businesses relying on AI-generated content—such as publishers or marketing teams—should prepare for more rigorous verification processes, even if the watermarks remain invisible to end users.
Sources
This story is an original summary drafted with AI by Lazyfounder from the reporting listed above and checked by automated validation. Facts are attributed to their original publishers; sections marked as analysis are Lazyfounder's. Where a source is in another language, facts were machine-translated and quotations are reported, not reproduced. Read the original coverage via the links, and see our AI policy and corrections policy.
About the author
Editor, Lazyfounder
Tarun Mottlia edits LazyFounders, covering Indian startups, funding rounds, AI and product launches. Every story on the site is AI-assisted and checked against its cited sources before publication.
More stories by Tarun MottliaGet the LazyFounder Brief
Startup, funding and AI news in a five-minute read. Join the early-access list.


